Base Scripte

💰 Business Value

Vollautomatischer YouTube-Video-Workflow: Video von EU-Parlament herunterladen, KI-Titel + Thumbnail generieren, auf YouTube hochladen, Datenbank synchronisieren — alles in einem Durchlauf.

⚙️ Funktionsweise

Die Kern-Mechanik des Projekts im Überblick.

💻 Code-Einblicke

Ein Blick unter die Haube — so ist das Projekt technisch umgesetzt:

EU-Parlament Video-Download & JSON-Parsing

response = requests.get(url)
html_text = response.text

json_pattern = re.compile(
    r'<script id="__NEXT_DATA__" type="application/json">(.*?)</script>')
match = re.search(json_pattern, html_text)

if match:
    data = json.loads(match.group(1))
    resolutions = (data.get("props", {})
        .get("pageProps", {})
        .get("mediaItem", {})
        .get("videos", [])[0]
        .get("resolutions", []))
    
    # Größte Auflösung finden und herunterladen
    filtered = [r for r in resolutions if r.get("resolution")]
    sorted_res = sorted(filtered, key=lambda r: r.get("size"), reverse=True)
    largest = sorted_res[0]
    
    # Download mit Streaming
    response = requests.get(largest.get("url"), stream=True)
    with open(f'{media_business_id}_fin.mp4', 'wb') as f:
        for chunk in response.iter_content(chunk_size=8192):
            if chunk:
                f.write(chunk)

OpenAI Titel-Generierung für YouTube

openai.api_key = config.openai_api_key

prompt = (f"Create a YouTube video title from this description: "
          f"{description}. The title should be engaging and "
          f"informative. Dont use thes charaters @, #, $, %, "
          f"&, *, +, =, |, /, {{, }}, [, ]. It should look "
          f"like someones Search request to youtube search and "
          f"has a maximum of 70 charaters")

settings = {
    "engine": "text-davinci-003",
    "prompt": prompt,
    "temperature": 0.7,
    "max_tokens": 1000,
    "n": 1,
    "stop": "."
}

response = openai.Completion.create(**settings)
for choice in response.choices:
    title = choice.text.strip()
    title = title.replace("engine", "").replace("bar", "")
    print(f"Title: {title}")

Thumbnail-Erstellung mit PIL & OpenCV

# Video-Datei öffnen und ersten Frame extrahieren
video = cv2.VideoCapture(f'{media_business_id}_fin.mp4')
ok, frame = video.read()
cv2.imwrite(f'{media_business_id}_alt.jpg', frame, 
            [int(cv2.IMWRITE_JPEG_QUALITY), 90])
video.release()

# Overlay-PNG auf Thumbnail legen
input_image = Image.open(f'{media_business_id}_alt.jpg').convert('RGBA')
overlay_image = Image.open('overlay_thumb.png').convert('RGBA')
result = Image.alpha_composite(input_image, overlay_image)

# Text mit Frage einzeichnen
draw = ImageDraw.Draw(result)
font = ImageFont.truetype('arialbd.ttf', 100)
text = quest  # OpenAI-generierte Frage
draw.rectangle((50, 800, box_width, box_height), fill=(255,255,255))
draw.text((60, 810), text, font=font, fill=(0, 0, 0))
result.convert('RGB').save(f'{media_business_id}thumb.jpg')

YouTube Upload mit API & DB-Update

request_body = {
    'snippet': {
        'categoryId': 25,
        'title': title,
        'description': description,
        'tags': tags
    },
    'status': {
        'privacyStatus': 'public',
        'selfDeclaredMadeForKids': False,
    },
    'notifySubscribers': False,
}

mediaFile = MediaFileUpload(VIDEO_FILE_NAME, chunksize=-1, 
                            resumable=True)
response_upload = googleAPI.videos().insert(
    part='snippet,status',
    body=request_body,
    media_body=mediaFile
).execute()

# Thumbnail setzen
googleAPI.thumbnails().set(
    videoId=response_upload.get('id'),
    media_body=MediaFileUpload(thumb)
).execute()

# Datenbank aktualisieren
sql = "UPDATE Videos SET ytID=%s, uploaded=1 WHERE ID=%s"
cursor.execute(sql, (response_upload.get('id'), mediaID))
conn.commit()

🚀 Gebaut mit Vibecoding

🎯 Strategische Erkenntnisse

Jedes Projekt liefert wertvolle Einsichten — technisch wie strategisch. Diese Learnings fließen direkt in Folgeprojekte ein.